Convolutional Neural Network Part Two: Building a CNN for Computer Vision

Показать описание

This episode is Part 2 of 2 of our exploration into the CIFAR-10 image dataset made available by the University of Toronto. In my last video on this topic, we explored how to download and explore the contents of the dataset in Python, so that we have an understanding of the images that we will be using for computer vision training and testing. In this episode, we're going to continue our script, build a convolutional neural network model in Python using Tensorflow/Keras, and then try to train our model on the image dataset to see if we can get any positive results. We'll use Keras, Tensorflow, Conv2D, MaxPooling2D, and many other features in order to accomplish our task. This will prepare us with a foundational model that we can configure for accuracy. We'll also monitor our CUDA/cudnn installation as it facilitates our GPU to take over processing from our CPU, saving hours of time in processing.

Related Videos:
Building a Database Application with ChatGPT
Convolutional Neural Network Part One: Preparing the CIFAR10 Dataset
Convolutional Neural Network Part Two: Building a CNN for Computer Vision
You are watching this one now!
Wildfire Time Lapse with Artificial Intelligence Prediction
Modify a Simple Machine Learning Script for Use with Your Data
How to Use Python to Build an Artificial Neural Network for Classification

Join me on Patreon!

Demo of my BZ RDP Cloaker:

Want the code from this video?

Interested in transforming your career or finding your next gig?

Want my team to do a project for you? Let's get to it!

Follow me on social media:

Get Microsoft Office including Access:

Got a YouTube Channel? I use TubeBuddy, it is awesome. Give it a try:

#python #datascience #dataengineering

Рекомендации по теме

Комментарии

Holly shit! I am trailing you all the time in my data engineering curiosities. I was just looking this material up!

interestingamerican

I bumped this up to 75%+ using a few more layers:

def cnn_model():
model = Sequential()
model.add(Conv2D(32, (3, 3), input_shape = (32, 32, 3), activation = 'relu'))
model.add(MaxPooling2D(strides=(2, 2), padding='same'))
model.add(Dropout(0.2))
model.add(Conv2D(64, (3, 3), activation = 'relu'))
model.add(Conv2D(64, (3, 3), activation = 'relu'))
model.add(MaxPooling2D(strides=(2, 2), padding='same'))
model.add(Dropout(0.2))
model.add(Conv2D(128, (3, 3), activation = 'relu'))
model.add(Conv2D(128, (3, 3), activation = 'relu'))
model.add(MaxPooling2D(strides=(2, 2), padding='same'))
model.add(Dropout(0.2))
model.add(Flatten())
model.add(Dense(128, activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes, activation = 'softmax'))
model.compile(loss='categorical_crossentropy', optimizer = 'SGD', metrics=['accuracy'])
return model

Are you able to get more accuracy? There are some cool research models on CIFAR-10 that get an unbelievable 99%+ accuracy!

Some notes:

- There are some preset algorithms that work awesome on CIFAR10, like the VGG16. I got over 85% very quickly using VGG16.
- Continue refining this model and then try using VGG16 to compare results.
- You can try wide vs. deep models. Wide uses more filters or neurons but fewer layers. Deep may use fewer neurons but more layers.
- Try using different configurations; epochs, optimizers, loss functions, randomizing training data can all make a difference.

seanmackenziedataengineering

How to open an Access database table and form and modify it in the form of an exe executable file

حجيةهيفاء-ضك

Convolutional Neural Network Part Two: Building a CNN for Computer Vision

Convolutional Neural Network Part Two: Building a CNN for Computer Vision

Ali Ghodsi, Lec 18: Convolutional neural network (Part 2)

Convolutional Neural Networks - Part 2

Convolutional Neural Networks (Part 2)

Pt 2: Deep Learning Convolution, Feature Hierarchy Abstraction, Convolutional Neural Networks

Convolutional neural networks, part 2

Introduction to Deep Learning - 10. Convolutional Neural Networks Part 2 (Summer 2020)

Introduction to Convolutional Neural Networks - Part II

[MXDL-12-02] Convolutional Neural Networks (CNN) [2/5] - 1D convolutional layer and 1D pooling layer

Convolutional Neural Networks [Part 2] - Filters

ISAAC: An Analog Convolutional Neural Network Accelerator (Part II)

ml5.js: What is a Convolutional Neural Network Part 2 - Max Pooling

Convolutional neural network (part 2): Convolution layer, pooling layer, and simple ConvNet Model

What is the Convolution in Convolutional Neural Networks? All About Neural Networks, Part 2

Convolutional neural network (part 2): ResNet, 1X1 Conv, Inception network, and object localization

Build a 2D convolutional neural network, part 2: Overview

Understanding Convolutional Neural Networks | Part 2 / 3 - Wonders of the world CNN with PyTorch

Deep Learning | Convolutional Neural Network - Part 2 | CloudxLab

HW for DL: Part 2b - Convolutional Neural Networks - Famous CNNs

Deep Learning Design Patterns Primer - Part 2 - Introduction to Convolutional Neural Networks

Convolutional Neural Networks: An Introduction - Part 02

CONVOLUTIONAL NEURAL NETWORK-PART 2

An Image Detecting Spreadsheet: Implementing Convolutional Neural Networks From Scratch Part 2

HW for DL: Part 2a - Convolutional Neural Networks